FINAL REPORT CONSTRUCTED WETLANDS FOR STORMWATER ... Wetland… · FINAL REPORT CONSTRUCTED...

FINAL REPORT

CONSTRUCTED WETLANDS FOR STORMWATER MANAGEMENT

Shaw L. Yu, Ph.D. Faculty Research Scientist

G. Michael Fitch Senior Research Scientist

T. Andrew Earles Graduate Research Assistant

INTRODUCTION

Wetland mitigation and stormwater management provisions in the 1987 CleanWater Act (CWA) significantly impact transportation agencies. CWA Section 404 stipulates thatwhen highway construction results in the displacement of natural wetlands, the highway agencyis required to create artificial wetlands to compensate for that loss. Section 402 directs the U.S.Environmental Protection Agency (EPA) to regulate stormwater runoff from certain areas underthe National Pollutant Discharge Elimination System (NPDES). Highway stormwater runoff,runoff from road construction sites with five or more disturbed acres, and runoff frommaintenance and storage facilities are subject to NPDES permit requirements.

In addition to the EPA regulations, the Virginia Department of Transportation(VDOT) must also comply with the Chesapeake Bay Preservation Act, Virginias StormwaterManagement Regulations, and State Erosion and Sediment Control Regulations. A commonrequirement of all of these stormwater regulations is the use of best management practices(BMP), such as detention ponds and infiltration for the control of runoff quantity and quality.

To date, VDOT has constructed over 220 wetlands and more than 350 stormwaterdetention basins. Wetland mitigation is a significant item in the VDOT road-building budget.Compliance with applicable stormwater regulations can add between ten and fifteen percent tothe cost of an average construction project. A potentially cost-effective approach to satisfyingwetland mitigation requirements and stormwater regulations is the use of mitigated wetlands asstormwater BMPs. It is believed that if a mitigated wetland site is properly engineered andmaintained, it will perform adequately as a stormwater BMP without jeopardizing its desiredwetland functions. It may also be possible to design a detention basin to include emergentwetland vegetation to enhance pollutant removal.

PURPOSE AND SCOPE

This study examined the potential for using mitigated wetlands as stormwaterBMPs in Virginia. The objectives of this research effort were:

1. To document water quality benefits of mitigated wetlands and vegetative detention basinsthrough monitoring of selected mitigated wetland sites and detention basins with emergentwetland vegetation. Monitoring focused on pollutant removal efficiency and stormwaterimpacts.

2. To develop a modeling framework for simulating the transport and fate of pollutants inwetland systems and to develop a link between a GIS and a watershed model for simulatingpollutant transport. The former provides a tool for predicting the effectiveness of constructedwetlands for water quality improvement; the latter allows planners to compare sitingalternatives for constructed wetlands and detention basins by looking at geographicalconstraints and anticipated runoff characteristics.

3. To develop a geographic information system (GIS) that would 1) serve as an inventory ofVDOTs constructed wetlands throughout the state; 2) become a selection tool for futureconstructed wetland site locations; and 3) serve as a data source for the hydrologic modelsused by VDOT.

The following report presents findings for each of these objectives in separate sections.Section I presents results of wetland monitoring, Section II describes wetland modeling, andSection III details GIS development.

SECTION I: WETLAND MONITORING

BACKGROUND

Constructed wetlands have been used for decades for the treatment of municipal andindustrial wastewater (Hammer, 1989 and Moshiri, 1993) and are considered to be more costeffective than advanced wastewater treatment systems. However, the use of wetlands to controlnonpoint source (NPS) pollution has only recently been investigated. A number of studiesconducted since the mid-1980s on the use of constructed wetlands for urban stormwatertreatment have suggested that constructed wetlands may improve stormwater runoff quality. Intheir study of the pollutant removal efficiency of a detention pond-wetland system receivingrunoff from a four-lane concrete roadway, Martin and Smoot (1986) found removal rates between41 and 73 percent for total solids, lead, and zinc. Athanas (1988), Munger, et al. (1995), Strecker,et al. (1990) have also reported significant removal of solids and metals in wetlands. Otherstudies have reported removal of nitrogen (10 to 50 percent) and phosphorus (16 to 70 percent)(Bautista and Geiger, 1993; Crumpton, 1995; Kappel, et al. 1985; Martin and Smoot, 1986;

2

Niswander and Mitsch,1995; Strecker et al., 1990). The complexity of nutrient cycling inwetland systems leads to a wide range of removal efficiencies for nitrogen and phosphorus.Depending on seasonal effects, vegetation type, and management practices, wetlands may serveas a source, sink, or transformer of nutrients (Raisin and Mitchell, 1995, Strecker, et al. 1990).However, well designed wetlands may attain long-term nutrient removals on the order of 25percent for total nitrogen and 45 percent for total phosphorus (Schueler, 1992).

The U.S. EPA encourages the use of constructed wetlands for nonpoint source pollutioncontrol, especially in agricultural areas (USEPA, 1992) and recognizes the need to combinewetland protection and nonpoint source pollution control strategies (Baker, 1993). A reportprepared for the Federal Highway Administration describes the applicability of constructedwetland technology for nonpoint source pollution from highways (Dorman, et al., 1988):

Artificial wetlands offer many more options for the management of highwayrunoff . . . the constructed wetland can be sized to accommodate a projected hydraulicload and to provide a specific residence time; constructed within the highway right-of-way, in median strips, in cloverleafs, or alongside the highway, and designed to facilitateoperations and maintenance.

Though data collected to document the performance of wetlands constructed forstormwater quality improvement are increasing (Yu, et al., 1996, 1997), most studies haveexamined wetlands designed specifically for water quality improvement rather than for thepurpose of mitigation.

METHODS

Site Selection, Preparation, and Description

There are over 220 mitigated wetland sites listed by VDOT as of 1997, most located inthe coastal region of Virginia. The VDOT Environmental Division and the research teamreviewed candidate sites to determine which would be best for sampling and monitoring.Candidate sites were selected based on the following criteria:

1. Site located on state or public property2. Stormwater runoff as main water source3. Vegetation well developed and aged at least 3 years4. Clearly defined inlet(s) and outlet(s)5. Accessibility by the research team

The research team visited candidate sites and compared them before making finalselections. The most common reason for elimination of a candidate site was the lack of well-defined inlets and outlets. Also, many sites were found unsuitable for sampling and monitoringbecause they had multiple inlets and outlets. In general, sites with more than three inlets andoutlets were not selected due to the limited number of automatic samplers available.

3

A total of eight sites, presented in Table 1, were ultimately chosen for monitoringover a three year study period. Locations of these sites are illustrated in Figure 1.

Table 1. Study Sites

Location YearsMonitored

Size ha (acre) Runoff Types Type of Wetland

Rio Hill Pond/ Rio HillShopping Center/Albemarle, VA

1995 0.28 (0.7) Parking Lot,Highway

Stormwater DetentionBasin with EmergentVegetation

Buster Pond/ RichmondVDOT District Office/Colonial Heights, VA

1996 0.04 (0.1) Office Complex,Auto Shop, GasPump Island

Natural Wetland

Route 637/ KingslandCreek Mitigation Area/Chesterfield, VA

1996 1.21 (3.0) Highway,Residential, Forest

Mitigated Wetland

Route 295/ Fort LeeMitigation Area/ Fort Lee,VA

1996-97 7.32 (18.1) Highway Mitigated Wetland

Route 288/ Mitigation Site14/ Chesterfield, VA

1996-97 2.02 (5.0) Highway Mitigated Wetland

Brooke Commuter RailParking Lot/ Brooke, VA

1996 0.08 (0.2) fordetention basin2.83 (7.0) forwetland

Parking Lot,Railway

Stormwater DetentionBasin with EmergentVegetation andMitigated Wetland

I - 64/ Bowers Hill/Chesapeake, VA

1996 0.70 (1.7) Highway Mitigated Wetland

Rt. 211/ Covington RiverMitigation Area/Rapahannock, VA

1997 0.40 (1.0) Agricultural,Highway

Mitigated Wetland

Figure 1. Stormwater sampling locations.

4

All sites were prepared for monitoring in accordance with the projects 1997 QualityAssurance/ Quality Control (QA/QC) Plan (provided in Appendix A). Figure 2 depicts aninflow monitoring station at the Rt. 288 site in Richmond. This station is a typical example of asite prepared for monitoring. Rainfall is measured by a tipping bucket rain gage, while flow ismeasured using a rectangular weir. An automatic sampler records rainfall and flowmeasurements and collects samples during runoff events. Similar monitoring stations wereestablished at all sites. Wetland delineations determined for the Rio Hill, Rt. 295, Rt. 288,Brooke, and Covington River sites are shown in Figure 3, along with their inlets and outlets.

Figure 2. Typical monitoring station (Rt. 288 infow)

Site Descriptions

Rio Hill

The Rio Hill detention basin (Figure 4) is a 0.28 ha (0.7 acre) impoundment serving a 30ha (75 acre) drainage area in Albemarle County. The box and raingage at the far end of the basinmonitor the basins outflow. Runoff to the basin is supplied from a shopping center with anextensive parking area and from a nearby intersection (average daily traffic (ADT) 33,000vehicles). Unvegetated open water area accounts for less than 5 percent of the wetland area. Thelower section of the basin usually has shallow standing water and is dominated by moderatelydense emergent vegetation including Juncus effusus (Soft Rush), Typha latifolia (Cattail),Euthamia graminifolia (Goldenrod), and Scirpus cyperinus (Wool Grass). Woody vegetation and

5

Figu

re 3

. W

etla

nd S

ite D

elin

eatio

ns

6

Figure 4. Rio Hill detention basin

Figure 5. Rt. 288 outlet zone

7

shrubs, primarily Gleditsia aquatica (Swamp Locust), Juniperus virginiana (Eastern Red Cedar),Albizzia julibrissin (Mimosa), and Pinus taeda (Loblolly Pine), are moderately dense in the highmarsh areas along the banks. Salix nigra (Black Willow) is dense along the main channel.

Buster Pond

The Buster Pond wetland is a small 0.04 ha (0.1 acre) wetland that receives runoff fromthe office complex, auto shop, and gas island at the VDOT Colonial Heights District Office.Salix spp. (Black Willow and Weeping Willow) are dominant in a small upland area in theproximity of the inlet; however, the vast majority of the wetland is permanently flooded andsupports a dense stand of Typha latifolia (Cattail). During the process of monitoring this siteduring 1995, it became apparent that this site was unsuitable for monitoring due to difficulty inmeasuring outflow. Since the outflow from the wetland merges into a pond, the outlet from thewetland is not well defined. Although monitoring is possible at the outlet of the pond, samplingat this point could not isolate the effect of the wetland alone on water quality.

Rt. 637 Kingsland Creek Mitigation Area

The Kingsland Creek Mitigation Area is a 1.21 ha (3.0 acre) constructed wetland inChesterfield designed as mitigation for impacts to palustrine emergent, scrub-shrub, and forestedwetlands resulting from road and bridge improvements. Woody species present at the siteinclude Acer rubrum (Red Maple), Betula nigra (River Birch), Liquidamber styraciflua (SweetGum), Juniperus virginiana (Eastern Red Cedar), Pinus taeda (Loblolly Pine), and Quercus alba(White Oak). Shrubs present include Vaccinium corymbosum (Highbush Blueberry) andViburnum dentatum (Arrowwood), while emergent species include Juncus spp. (Rush), Carexspp. (Sedge), and Polygonum spp. (Tear Thumb). Monitoring at this site was discontinued due toa leaking concrete culvert at the inlet. Flow measurement and sampling equipment was set up atthe end of the culvert at the point of discharge into the wetland; however, gaps between thesections of the culvert allowed water to leak from the culvert into the ground prior to themonitoring point. While this water eventually reached the wetland by flowing beneath theculvert, there was no effective way to monitor it.

Rt. 295 Fort Lee Mitigation Area

This 7.32 ha (18.1 acre) site in Fort Lee was the largest site monitored in thestudy. Although monitoring of all inflows and outflows at this site proved to be beyond theresources of the project, the researchers studied the runoff entering the mitigation area from oneinflow for two years in order to characterize it. Land use (primarily interstate highway Rt. 295and associated right-of-way) was similar for all inflows into the wetland. In addition to thisinflow station, in 1997 some outflow measurements were collected in the main outflow channel.The majority of the wetland is semi-permanently flooded and supports a variety of emergentvegetation that is dominated by Scirpus cyperinus (Wool Grass), Juncus spp. (Rush), and Typhalatifolia (Cattail).

8

Rt. 288 Mitigation Area 14

The Rt. 288 mitigation site is a 2.02 ha (5 acre) mitigated wetland located in the medianof a four-lane highway with a 50,000 vehicle ADT in Chesterfield. The site is characterized by acombination of wet meadow, fresh marsh, and tree swamp area, with a large open water zonenear the outlet (approximately 25 percent of the site). Although some dry areas exist, soilconditions are mainly saturated, evidenced by shallow standing water (2.5-10 cm) covering mostof the site. Salix nigra (Black Willow), Scirpus cyperinus (Wool Grass), Juncus effusus (SoftRush), and Typha latifolia (Cattail) were prevalent in areas with shallow standing water. Lemnaspp. (Duckweed) was prevalent in open water areas, while Pinus taeda (Loblolly Pine) wasprevalent in the higher marsh area. Figure 5 shows the outlet zone of the Rt. 288 wetland.

Brooke Detention Basin and Wetland

The Brooke site consists of a 0.08 ha (0.2 acre) emergent detention pond and a 2.83 ha (7acre) mitigated wetland in series. The site receives stormwater runoff from a commuter parkinglot, a grassed area, and a railway. Conditions range from permanently flooded regions where deep(up to 1 m) pools exist to intermittently flooded regions where surface water is present duringstorm events and near-saturated to saturated soil conditions prevail during dry weather. The sitehas approximately 0.4 ha (14 percent of total area) of open water. The detention basin isintermittently flooded with water levels rising as high as 2 m during large storm events. Like therest of the site, the basins soil is usually at or near saturation during dry periods. Vegetationdensity is moderate to dense in all but the open water area. Scirpus cyperinus (Wool Grass),Typha latifolia (Cattail), and Juncus effusus (Soft Rush) are the dominant emergent species andSalix nigra (Black Willow) is dense along the main channel of the wetland. Primary species inthe detention basin are Scirpus cyperinus (Wool Grass), Typha latifolia (Cattail), Solidago spp.(Goldenrod), and Juncus effusus (Soft Rush).

I-64 Bowers Hill Mitigation Site

The Bowers Hill site is a 0.70 ha (1.7 acre) mitigated wetland in Chesapeakesurrounded on all sides by highways. The primary sources of runoff for the wetland are theeastbound lanes of I-64 and an exit ramp from the westbound lane. While some dry areas exist,soil conditions are mainly saturated, evidenced by shallow standing water covering much of thesite. Dense vegetation at this site includes Typha latifolia (Cattail), Arundinaria gigantea (GiantCane Grass), and woody species including Acer rubrum (Red Maple), Quercus michauxii(Swamp Chestnut Oak) and Quercus bicolor (Water Oak).

Rt. 211 Covington River Mitigation Area

The Covington River wetland is an approximately 0.4 ha (1 acre) mitigation area, locatedin an agricultural area of Rappahannock County. Soil conditions range from extremely dry onthe western side of the wetland in a high marsh area to saturated conditions with shallow pools of

9

standing water in the low marsh covering the remainder of the site. Unvegetated open water areaaccounts for less than 5 percent of the wetland area. The upper marsh is dominated byAndromeda glaucophylla (Bog Rosemary), while common low marsh species observed includeJuncus effusus (Soft Rush), Impatiens capensis (Jewel Weed), Aster puniceus (Purple Aster), andTypha latifolia (Cattail). Salix nigra (Black Willow) is dense along the banks of the channelwhile Acer rubrum (Red Maple), and Alnus serrulata (Common Alder) are common throughoutthe wetland. Figure 6 depicts the main channel of the Covington River mitigation area. Thoughthis photograph was taken prior to the start of the growing season, the abundance of dead plantmaterial in the picture is indicative of the productivity of this site.

Figure 6. Covington River Mitigation Area

Sample Analysis

Analytical parameters for analysis were selected based on the objectives and resources ofthe project. Since a primary objective of the project was to monitor highway runoff enteringconstructed wetlands, it was important that constituents chosen for analysis adequatelycharacterize stormwater runoff. Table 2 lists analytical parameters recommended by the Nation-Wide Urban Runoff Program (NURP) to adequately characterize urban runoff. Parameterschosen were total suspended solids (TSS), chemical oxygen demand (COD), total phosphorus(TP), orthophosphate (OP), and zinc (Zn). In addition, the researcher performed limited analysisof total nitrogen (TN), nitrate (NO

3-N), and fecal coliform (FC) for the Covington River site.

Samples were collected using automatic samplers at monitoring stations during stormevents. The research team retrieved, chilled, and preserved the samples as soon as possible aftercollection. Samples were analyzed at the University of Virginia Stormwater Laboratory inaccordance with the QA/QC plan in Appendix A. The University of Virginia Stormwater

10

Laboratory utilized the spectrophotometric, U.S.EPA-approved Zincon method (Hach Co., 1991)to perform most zinc analyses. Analysis of samples with very low levels of zinc was onlypossible by atomic absorption spectrometry, however. The Aqua Air Laboratory inCharlottesville, Virginia performed this analysis.

Three methods were used to examine pollutant removal in the constructed wetlands. Amass balance was used to determine mass removal efficiency during storm events based onvolumes of inflow and outflow and event mean concentrations (EMC) from the compositesamples. Mass removal efficiency (MRE) was calculated as:

The researchers also used the determination of event mean concentration (EMC) removalefficiency to assess pollution removal. Event mean concentrations were determined from analysisof flow-weighted composite samples or from flow-weighting of analytical results from analysisof discrete samples. EMC removal was calculated as:

The third method used was the calculation of the sum of the loads (SOL) efficiency. SOLefficiency is calculated based on mass entering and leaving the wetland over all stormscompletely monitored (a storm was considered to have been completely monitored whenstormwater samples and flow measurements were collected at every monitoring station at thesite). SOL removal efficiency was calculated as:

Of the three methods, the EMC reduction method is the most conservative estimate ofremoval for a single event, since it does not account for the storage of water. The SOL method,on the other hand, provides a calculation of removal over a longer time period than an individualevent. It should be noted that in some instances, the calculated MRE or SOL removal will belower than the EMC reduction. EMC removal efficiencies are greater than MRE when rain(assumed to have minimal pollutant concentrations) falling directly onto a wetland with littlestorage available results in additional outflow. The contribution of direct rainfall can be quitesignificant, especially for the larger sites. For most sites EMC, MRE, and SOL removalefficiencies were similar; however, large variation between these three measures was observed at

EMC Efficiency (%) = outlet EMCinlet EMC

1001FHGIKJ

SOL Efficiency (%) =

Volume in (i) Concentration in (i) Volume out (i) Concentration out (i)

Volume in (i) Concentration in (i)

i=1

# storms

i= 1

#storms

# storms

F

H

GGGG

I

K

JJJJ

i 1

100

11

MRE Volume in Concentration in) (Volume out concentration out)Volume in Concentration in)

100(%) ((

the Brooke site. This variation is attributable to the small number of storms monitored at this siteand the large amount of storage for the storms that were monitored.

Vegetation and Wildlife Monitoring

Beginning in 1996, surveys of vegetation were conducted at the Rt. 288, Brooke,Rio Hill, and Covington River sites. The research team conducted an initial survey of each siteto identify the species present. After this inventory was performed, square meter plot counts ofvegetation were conducted throughout the wetland. Density of vegetation was noted on aqualitative scale ranging from sparse to abundant. A qualitative assessment of vegetative healthwas also made. Observations of wildlife present, whether from visual identification or fromindirect evidence such as animal tracks, were made every time the sites were visited.

RESULTS

Between 1995 and 1997 fifty-nine storm events were monitored at the eight sites. Thirty-nine of these events were considered completely monitored. For sites where inflows andoutflows were monitored (Rio Hill, Rt. 288, Rt. 460, Brooke, and Covington River) thesecompletely monitored events yielded data necessary for EMC, MRE, and SOL analysis. Itshould be noted that inflow 3 at Rt. 288 was not monitored until 1997, since the contribution ofthis source is minor compared to inlets 1 and 2. Table 3 lists storms monitored by site.

Table 2. Analytical Parameters Recommended by NURP to Adequately Characterize Urban Runoff(USEPA 1991)

Conventional Parameters pHTotal Suspended SolidsBiological Oxygen DemandChemical Oxygen DemandSettleable SolidsTemperature

Metals CopperLeadZinc

Nutrients Total PhosphorusSoluble PhosphorusTotal Kjeldahl NitrogenNitrate/Nitrite Nitrogen

Biological Parameters Fecal Coliform

12

Precipitation and Flow

Mean precipitation data for storms monitored at the study sites are shown in Figure 7.Only storms that produced runoff are indicated in the range of precipitation. For sites with highlyimpervious drainage areas such as the Rio Hill and Brooke sites, as little as 5 mm of rainfallproduced sufficient rainfall to trigger samplers. Figure 8, constructed from data collected duringa storm event at the Rt. 288 site on November 3, 1995, illustrates typical flow and precipitationdata.

Flood control is an important hydrologic function of wetlands. This function is especiallycritical for mitigated wetland sites located in developed or paved areas where there is morerunoff due to increased imperviousness. Results from six storm events monitored at Route 288,five events monitored at Brooke, and four events monitored at the Covington River site indicatethat these wetlands significantly reduced peak flows during storm events. An average peakreduction of 40 percent was observed at the Rt. 288 site for rainfall events ranging from 18.3 mm(0.72 in) to 51.6 mm (2.03 in). The most intense storm monitored at Route 288, with 40.4 mm(1.59 in) of rain in 4 hours, resulted in a peak reduction of 58 percent. At the Brooke site,inflows and outflows were monitored for rainfall events ranging from 3.8 mm (0.15 in) to 47.2mm (1.86 in). At the Brooke site, the detention basin is a major factor in peak reduction,reducing peak inflows an average of 83 percent. Data from the Brooke wetland indicate averagepeak reductions of 46 percent, resulting in an average peak reduction of nearly 90 percent for thesystem. The most intense event monitored, with 47.24 mm (1.86 in) of rain in 50 minutes,resulted in a peak reduction of 93 percent for the entire system. An average peak reduction of 48percent was observed at the Covington River site for rainfall events ranging from 2.0 mm (0.08in) to 28.5 mm (1.12 in).

Water Quality

Total Suspended Solids (TSS)

TSS inflow concentrations varied widely between study sites. They ranged from less than8 mg/L at the Rt. 295 site, where flow passes over a grassed embankment prior to collection, towell over 300 mg/L at the Rio Hill site, where ongoing construction and a highly impervious

Table 3. Storm Events Monitored

Site Number ofStations

Number of Events Number of Completely MonitoredEvents

Rio Hill 3 5 5Buster Pond 2 2 2Rt. 637 1 2 2I-295 1 12 11Rt. 288 4 13 7Brooke 3 13 3 for basin, 2 for wetland, 1 overallRt. 460 3 7 2Covington River 2 5 3

13

Figure 7. Mean precipitation and range for storms monitored

14

Figure 8. Typical flow and precipitation data collected at monitoring station

Figure 9. Mean TSS inflow concentration at study sites

15

surface contributed to high levels. Figure 9 illustrates mean TSS inflow concentrations at studysites, with error bars showing one standard deviation. In general, greater variability wasobserved for higher mean concentrations.

EMC reduction of TSS is shown in Figure 10 and is presented in Table 4 along with MREand SOL removal (standard deviations are noted parenthetically).

Figure 10. Mean EMC reduction of TSS

Table 4. Removal Efficiency for Total Suspended Solids (TSS)

Site EMC Reduction % and(standard deviation %)

MRE % and(standard deviation %)

SOL Removal%

Rt. 288 56.96 (16.08) 61.17 (15.75) 52.02Covington River 29.02 (12.19) 68.29 (30.77) 62.45Rio Hill -1.32 (29.02) 4.98 (24.13) 30.10Rt. 460 54.35 (6.76) 40.36 (12.26) 48.67Brooke Detention Basin 52.38 (26.94) 67.48 (4.43) 65.68Brooke Wetland -66.75 (157.02) 60.66 ( ----- ) 60.66Brooke Overall -----1 ----- -----Buster Pond 8.71 (118.40) ----- -----1 Indicates insufficient data

16

Average EMC reduction as high as 57 percent (with even higher MRE and SOL reductions) wasobserved at the Route 288 site. Analysis of variance for the Rt. 288 site indicated a significant (= 0.05) difference between mean TSS concentration in the outflow and at least one inflow. Theincrease in concentration observed at the Brook wetland should be viewed with two factors inmind: 1) the extremely high standard deviation and small sample size (two events with 44percent and 178 percent changes, respectively); and 2) positive MRE and SOL removal,indicating the effect of retention of stormwater in the wetland. Greater variability in inflowconcentrations tended to produce greater variability in removal (as evidenced by Rio Hill).Nonetheless, SOL calculations indicated removal of TSS for all sites.

Total Phosphorus (TP)

Figure 11 illustrates mean TP inflow concentrations at study sites with error bars showingone standard deviation. As with TSS, greater variability was observed for higher meanconcentrations. Mean inflow concentrations range from less than 1 mg/L at the Brooke site tonearly 3 mg/L at Rio Hill. Mean TP concentrations are strongly correlated (0.88) with mean TSSconcentrations, indicating that a significant portion of influent TP may be bound to solids.

Figure 11. Mean TP inflow concentration at study sites

EMC reduction of TP is shown in Figure 12 and is presented in Table 5 along with MREand SOL removal (standard deviations are noted parenthetically). EMC reduction ranged from 19 percent for Rt. 460, to nearly 70 percent for the Rt. 288 site. The sites with the best removalof TP, Rt. 288 and Covington River, showed far less variation (standard deviation < 10 percentfor EMC reduction) than the sites that did not perform as well. Analysis of variance for the Rt.288 site indicated a significant ( = 0.05) difference between mean TP concentration in theoutflow and at least one inflow. Poor TP removal reported for the Rt. 460 site was based on only

17

one storm, and data for removal at Brooke were also limited. The sites with the most extensivedocumentation (Rio Hill, Rt. 288, Covington River, and the Brooke Basin) indicated SOL TPremoval greater than 20 percent.

Figure 12. Mean EMC reduction of TP

Orthophosphate (OP)

Figure 13 shows mean inflow concentrations for OP with error bars indicating onestandard deviation. As with other parameters, larger standard deviations were generallyassociated with higher mean concentrations. Average inflow concentrations were less than 1 mg/L for the most pervious sites (Rio Hill and Brooke). They were about 2 mg/L for sites whereinflow passed over grassed embankments prior to entering the inlet channel of the wetland (Rt.288 and Rt. 295).

Table 5. Removal Efficiency for Total Phosphorus (TP)



SOL Removal%

Rt. 288 68.61 (8.67) 68.92 (17.67) 68.09Covington River 29.46 (8.37) 71.22 (23.79) 67.36Rio Hill 14.86 (29.95) 18.89 (25.42) 27.46Rt. 460 -19.11 (-----1) -61.35 (-----) -61.35Brooke Detention Basin -4.22 (37.25) 4.27 (34.32) 22.85Brooke Wetland 45.83 (-----) ----- -----Brooke Overall 66.23 ----- -----Buster Pond 2.19 ----- -----1 Indicates insufficient data

18

Figure 14 illustrates EMC reduction of OP. Table 6 presents MRE and SOL removal(standard deviations are noted parenthetically).

Figure 13. Mean OP inflow concentration at study sites

Figure 14. Mean EMC reduction of OP

19

Excellent removal was achieved at the Rt. 288 site, with significant mass removal also at theCovington River site. Analysis of variance for the Rt. 288 site indicated a significant ( = 0.05)difference between mean OP concentration in the outflow and at least one inflow. OP removalfollowed a trend similar to TP removal, with the Brooke detention basin and the Rt. 460 siteshowing poorest performances. As with TP data, OP data for these sites were limited, resultingin large standard deviations for removal. On a SOL basis, there was positive removal of OP forall but the Rt. 460 site. Observed increases in concentration are likely attributable to release ofbound phosphorus from sediments due to decomposition and hydrolysis of organic phosphorus.

Chemical Oxygen Demand (COD)

Figure 15 illustrates mean COD inflow concentrations at study sites with error barsshowing one standard deviation.

Figure 15. Mean COD inflow concentration at study sites

Table 6. Removal Efficiency for Orthophosphates (OP)



SOL Removal%

Rt. 288 81.50 (6.78) 81.6 (14.26) 82.46Covington River 15.38 (-----1) 32.21 (-----) 32.21Rio Hill -8.00 (20.49) -4.59 (21.87) 0.67Rt. 460 -38.41 (22.65) -80.92 (39.98) -56.24Brooke Detention Basin -19.34 (54.23) 11.04 (81.04) 45.74Brooke Wetland 0.45 (-----) 1.12 (-----) 1.46Brooke Overall ----- ----- -----Buster Pond -7.33 (-----) ----- -----1 Indicates insufficient data

20

Average concentrations ranged from slightly over 20 mg/L at Buster Pond to over 80 mg/L forthe average inflow at Rt. 460. Variability in event mean inflow concentration tended to increaseas the average concentration increased. With the exception of the Rt. 460 inlet 1, average CODinflow concentrations did not vary as drastically between sites as other parameters monitored.

EMC reduction of COD is shown in Figure 16 and is presented in Table 7 along withMRE and SOL removal (standard deviations are noted parenthetically). As with otherparameters monitored, positive removal was achieved by the Rt. 288 site (greater than 20percent) and the Covington River site (greater than 50 percent). Although the Brooke wetlandhad a high percent increase in EMC concentration and caused increased COD on a mass basis,the Brooke site overall had a significant removal of COD. This removal was attributable togreater than 60 percent removal of COD by the detention basin. As with TP and OP, the Rt. 460wetland performed poorly based on some events sampled. SOL removal was positive for thissite, however, indicating that extremely poor performances for individual events may haveskewed EMC reduction and MRE averages.

Figure 16. Mean EMC reduction of COD

21

Zinc (Zn)

Figure 17 illustrates mean Zn inflow concentrations at study sites with error bars showingone standard deviation. Correlation between TSS average inflow concentration and Zn averageinflow concentration was poor (-0.40), suggesting that a large fraction of Zn entering the studiedwetlands was in a dissolved form. As with other parameters monitored, variability of Zn inflowconcentration tended to increase as the average inflow concentration of Zn increased. An initialtest for Zn at the Covington River site was below the detection limit of the atomic absorptionmethod (.05 mg/L), so further testing for Zn at this site was discontinued.

Table 7. Removal Efficiency for Chemical Oxygen Demand (COD)



SOL Removal%

Rt. 288 23.24 (11.87) 22.18 (28.86) 24.23Covington River 50.33 (22.98) 74.63 (74.63) 62.19Rio Hill -31.59 (42.67) -24.88 (29.13) -22.76Rt. 460 -5.88 (40.85) -32.29 (61.19) 2.21Brooke Detention Basin 58.51 (24.85) 67.19 (16.05) 71.73Brooke Wetland -161.38 (-----1) -145.86 (-----) -132.64Brooke Overall 69.66 (-----) ----- -----Buster Pond -6.82 (-----) ----- -----1 Indicates insufficient data

Figure 17. Mean Zn inflow concentration at study sites

22

EMC reduction of Zn is shown in Figure 18 and is presented in Table 8 along with MREand SOL removal (standard deviations are noted parenthetically). Consistent removal in therange of 40 percent was found at the Rt. 288 site, while average removals at other sites rangedfrom around 20 percent at Rio Hill to more than 80 percent for the Brooke detention basin.Though concentration of Zn increased in the Brooke wetland, mass was removed as indicated byMRE and SOL removal. No change in Zn concentration was observed at the Buster Pond site.

Figure 18. Mean EMC reduction of Zn

Table 8. Removal Efficiency for Zinc (Zn)



SOL Removal%

Rt. 288 43.01 (6.31) 48.12 (22.59) 31.63Covington River -----1 ----- -----Rio Hill 24.23 (13.51) 26.22 (16.85) 29.47Rt. 460 53.93 (-----) 42.55 (-----) 42.55Brooke Detention Basin 86.79 (-----) ----- -----Brooke Wetland -33.33 (-----) 36.28 (-----) 62.35Brooke Overall ----- ----- -----Buster Pond 0.00 (-----) ----- -----

23

Other Parameters

In addition to monitoring of TSS, TP, OP, COD, and Zn, samples from the CovingtonRiver site were analyzed for total nitrogen (TN), nitrate (NO

3-N), and fecal coliform (FC).

Influent concentrations of 3.8 mg/L, 1.6 mg/L, and 8000 cfu/100mL were reduced to outflowconcentrations of 3.1 mg/L, 0.73 mg/L, and 2000 cfu/100mL for TN, NO

3-N, and FC,

respectively, resulting in EMC reductions of 18.42 percent for TN, 54.38 percent for NO3-N,

and 75.00 percent for FC.

Vegetation and Wildlife

The research team collected vegetation data at Rio Hill, Rt. 288, Brooke, and CovingtonRiver to examine the diversity and abundance of vegetation present and to document any appar-ent impacts due to stormwater runoff. Initially, the team conducted a survey of species present.Subsequently, square meter plots were studied to assess the relative abundance of wetland plants.The researchers noted vegetation density on a qualitative scale ranging from sparse to abundant.During each site visit (approximately once a week), the team also recorded any observations ofwildlife present. Observations included visual identification and/or indirect evidence such asanimal tracks.

Figures 19a-d illustrate the distribution of dominant plant species at each site wherevegetation was monitored. No record for initial planting was available for the Rio Hill site;however, the abundance of species observed (greater than 20) is far greater than that of a typicalplanting plan. Density of vegetation at the Rio Hill site was moderate with scattered stands thatwere very dense. Only a small, dry section along the southeastern bank was sparsely vegetated.Mitigation plans at the other three sites specified the planting of a variety of emergent plants,shrubs, and woody species. The plans specified in-kind replacement of any species not survivingafter the first year. Plantings took place in 1991, 1992, and 1993 for the Brooke, Rt. 288, andCovington River sites, respectively. No planting was performed for the Brooke detention basin.Comparisons of species planted and species observed are presented in Tables 9 11. Density ofvegetation was moderate to very dense at all three of these sites, with the greatest density at Rt.288.

24

Table 9. Planted and Observed Vegetation at Route 288 Site

Species at Route 288 SitePlanted 1992 Observed 1996

Emergent, FloatingAquatic Vegetation, andWildflowers

Polygonum pennsylvanicum (Giantsmartweed)Echinochloa crusgalli (JapeneseMillet)

Scirpus cyperinus (Wool grass)Typha latifolia (Broad leaf cattail)Juncus effusus (Soft rush)Rhynchospora capitellata (Small-headed beak rush)Eleocharis rostellata (Beaked spikerush)Scirpus atrovirens (Green bulrush)Carex scoparia (Broom sedge)Carex lurida (Lurid sedge)Cicuta maculata (Water hemlock)Ludwigia alternifolia (Seedbox)Sphagnum magellanicum (Sphagnummoss)Polygonum pennsylvanicum (Giantsmartweed)Lemna spp. (Duckweed)

Shrubs Ilex verticillata (CommonWinterberry)Cephalanthus occidentalis(Buttonbush)Sambucas canadensis (AmericanElder)Viburnum dentatum (SouthernArrowwood)

Alnus serrulata (Common Alder)Cephalanthus occidentalis(Buttonbush)

Woody Species Betula nigra (River Birch)Salix nigra (Black willow)Fraxinus pennsylvanica (Green Ash)Nyssa sylvatica (Black gum)

Betula nigra (River Birch)Salix nigra (Black willow)Juniperous virginiana (Eastern RedCedar)Pinus taeda (Loblolly Pine)Acer rubrum (Red Maple)Fraxinus pennsylvanica (Green Ash)

27

Table 10. Planted and Observed Vegetation at Brooke Site

Species at Brooke SitePlanted 1991 Observed 1996


Peltra virginica (Arrow Arum)Saururus cernuus (Lizards Tail)Leersia oryzoides (Rice cutgrass)

Peltra virginica (Arrow Arum)Saururus cernuus (Lizards Tail)Leersia oryzoides (Rice cutgrass)Cicuta maculata (Water hemlock)Juncus effusus (Soft rush)Typha latifolia (Broad leaf cattail)Carex scoparia (Broom sedge)Carex lurida (Lurid sedge)Lemna spp. (Duckweed)Eupatorium maculatum (Joe-Pye-Weed)Pluchea camphorata (StinkingMarsh-Fleabane)Sphagnum magellanicum (Sphagnummoss)Erigeron annus (Daisy Fleabane)Solidago spp. (Goldenrod)Hypericum spp. (St. Johns Wort)

Shrubs Alnus serrulata (Common Alder)Cephalanthus occidentalis(Buttonbush)Sambucus canadensis (CommonElder)

Alnus serrulata (Common Alder)Cephalanthus occidentalis(Buttonbush)Sambucus canadensis (CommonElder)

Woody Species Betula nigra (River Birch)Liquidambar styraciflua (Sweetgum)Acer rubrum (Red Maple)

Betula nigra (River Birch)Liquidambar styraciflua (Sweetgum)Acer rubrum (Red Maple)Fraxinus pennsylvanica (Green Ash)Pinus taeda (Loblolly Pine)

28

Table 11. Planted and Observed Vegetation at Covington River Site

Species at Covington River SitePlanted 1993 Observed 1997


Juncus effusus (Soft rush)Typha latifolia (Broad leaf cattail)Scirpus validus (Softstem bulrush)Scirpus pungens (Common three-square)

Juncus effusus (Soft rush)Scirpus validus (Softstem bulrush)Scirpus pungens (Common three-square)Scirpus cyperinus (Wool grass)Impatiens capensis (Jewel weed)Polygonum sagittatum (Arrow leafedtearthumb)Polygonum punctatum (Watersmartweed)Carex lurida (Lurid sedge)Euthamia graminifolia (Goldenrod)Mentha spicata (Spearmint)Typha latifolia (Broad leaf cattail)Vernonia noveboracensis (New Yorkironweed)Eupatorium maculatum (Joe-Pye-Weed)Liliceae spp. (Lily family vegetation)Aster puniceus (Swamp aster)Cuscuta grovovii (Common dodder)Verbesina alternifolia (Wingstem)Pluchea camphorata (StinkingMarsh-Fleabane)Galium tinctorium (Dye bedstraw)Elatine americana (Americanwaterwort)

Shrubs Alnus serrulata (Common Alder)Viburnum spp. (Arrowwood)Ilex verticillata (Winterberry)

Alnus serrulata (Common Alder)Viburnum spp. (Arrowwood)Ilex verticillata (Winterberry)

Woody Species Betula nigra (River Birch)Liquidambar styraciflua (Sweetgum)Acer rubrum (Red Maple)Quercus spp (Oak)Fraxinus pennsylvanica (Green Ash)Salix nigra (Black Willow)

Betula nigra (River Birch)Liquidambar styraciflua (Sweetgum)Acer rubrum (Red Maple)Quercus spp (Oak)Fraxinus pennsylvanica (Green Ash)Salix nigra (Black Willow)

29

The number of plant species observed at each site is one measure of diversity. Therelative abundance of these various species is also important as well. While the scale of the one-meter plots is too small to accurately evaluate the composition of woody species in an area, suchplots reveal considerable information on emergent species diversity. Figures 20-22 show therelative abundance of species in the 288, Brooke, and Covington River wetlands, respectively.These figures are based on composites of all meter plots conducted at the site as to provide anoverall composition of the wetland. They do not include woody species or floating aquatic plantsthat cannot be easily counted individually such as Lemna (Duckweed). Both Rt. 288 and Brookehave strong presence of Typha latifolia (Cattail) and Juncus effusus (Soft Rush); however, neitherspecies presence is overwhelming. Since the Covington River site was not monitored until1997, the amount of vegetation data collected at this site were limited. Species composition failsto reflect some significant populations present, including a dense Typha latifolia stand.

Figure 20. Composite composition (%) of vegetation at Rt. 288

30

Figure 21. Composite composition (%) of vegetation at Brooke

Figure 22. Composite composition (%) of vegetation at Covington River

31

The research team observed a variety of wildlife at sites during field visits. Since visits tothe sites were almost exclusively during daylight hours (most often in mid-afternoon) nocturnalspecies and species that prefer cooler periods of the day were not likely to be observed. Themost extensive observation of wildlife was at the Rt. 288, Brooke, and Covington River sites.Wildlife observed at these sites include beaver, muskrat, field mice, frogs, turtles, ducks, snails,and a variety of snakes. Numerous birds, including red winged black birds and a blue heron werealso observed. Deer, field mice, a variety of snakes, frogs, and turtles have been observed at theCovington River and Rt. 295 sites as well.

While the presence of wildlife at the wetland sites is a promising indicator of successfulhabitat creation, the activity of beaver at the Route 288 site presented problems for monitoringthe site (Figure 23). From late 1996 to June 1997, the beavers in the wetland constructed a largedam at the outflow from the wetland. While the additional storage created by the dam greatlyincreased stormwater retention, the dam impeded flow measurement and was detrimental tosome of the less water-tolerant vegetative species. Pinus taeda (Loblolly Pine) especiallyexperienced an increased hydroperiod from the damming. The research team constructed andinstalled a pond leveler to subvert the beaver dam. Monitoring since August 1997 indicatesexcellent performance of the pond leveler.

Figure 23. Beaver Dam at Rt. 288 Outflow

32

DISCUSSION

To utilize mitigated wetlands as stormwater BMPs, two sets of priorities must beaddressed. First and foremost, the mitigated wetland must replace functions lost in thedisplacement of the natural wetland. Second, the wetland must effectively control the quantityand quality of stormwater runoff. The diverse vegetation and habitat observed at the Route 288,Brooke, and Covington River mitigation sites, coupled with pollutant removal efficiencies andpeak attenuation comparable to conventional BMPs such as detention ponds (Yu et al. 1997),illustrate the ability of well-designed mitigation sites to serve both priorities.

All sites monitored supported apparently healthy wetland vegetation. Speciesidentification at Rt. 288, Brooke, Rio Hill, and Covington River indicated more than 20 speciesat each site. Observations at these and other study sites indicated moderately dense to very densevegetation despite the input of highway runoff as a primary water source. A comparison of theRt. 288 and Brooke sites reveals that mean inflow concentrations into the Rt. 288 wetland weretwo to eight times higher than mean inflows into the Brooke wetland. Despite these significantdifferences in pollutant inputs, similarities exist in vegetative diversity and wildlife habitat. Bothwetlands support over 20 vegetative species, with significant populations of Juncus effusus (SoftRush) and Typha latifolia (Cattail). The balance of emergent vegetation, shrubs, and woodyvegetation is similar at the Rt. 288 and Brooke wetlands, as well. Both wetlands provide habitatfor a variety of wildlife including birds, reptiles, and small mammals. The Brooke detentionbasin and Rio Hill basin, which receive runoff with pollutant concentrations comparable to theRt. 288 wetland inflow, also support dense emergent vegetation with no apparent impacts fromthe pollutant loading.

Wildlife observations at these sites demonstrate that a mitigated wetland that is acting asa BMP can also provide habitat for many species. Despite the physical barrier and noiseassociated with its location in the median of a four-lane highway, the Rt. 288 site appears toprovide a habitat for a range of animals similar to that of the Brooke and Covington River sites,both of which are located adjacent to streams in rural areas. Frequency of observation wassimilar at these three sites, with observations of wildlife common during even the shortest visitsto the sites. The large size of the Rt. 288 wetland and the presence of a significant portion ofopen water likely contribute to its success as a habitat despite its adverse location.

The Rt. 288 and Covington River mitigation sites illustrate the ability of a well- designedmitigated wetland to also serve as a stormwater BMP. Both sites achieved removal of allpollutants monitored with mean EMC reductions as high as 57 percent for TSS, 68 percent forTP, 81 percent for OP, 50 percent for COD, and 43 percent for Zn. While the performance ofthese sites is comparable to conventional BMPs, other sites monitored tended to produce mixedresults, with a great deal of variability in removal from storm to storm. This variability is due inpart to the small number of events sampled for some sites and also to wide variation in stormcharacteristics (i.e. a rainfall range of 10.2 to 98.6 mm for the Rt. 460 site). It should be notedthat with the exception of the Rt. 460 site, positive EMC reductions were within one standarddeviation of the mean EMC reduction for all sites monitored.

33

Differences in removal efficiencies for the sites monitored are likely attributable todifferences in key design parameters, including the configuration of inlets and outlets, the lengthto width ratio, and (consequently) residence time. Both the Rt. 288 and Covington River sitesare fairly linear with inlets adequately separated from the outlet as reflected by their length towidth ratios (4:1 (4.0) and 5:3 (1.67), respectively). Outflow at both sites is controlled bycontracted rectangular weirs. A primary difference between these two sites, however, is thenature of flow through the wetland. Much of the flow at Rt. 288 is shallow flow throughvegetation, while the flow at Covington River is very channelized with minimal obstruction fromvegetation. Some short-circuiting is suspected at Rt. 288 due to the proximity of inlet 3 to theoutlet; however, the magnitude of flow at inflow 3 is small relative to the contributions of theother inlets. Resulting average residence times of 27.9 hours and 8.5 hours for Rt. 288 andCovington River, respectively, reflect these differences in flow characteristics.

The Rio Hill detention basin, on the other hand, has several inlets spread out around thedetention basin. The position of the inlets with respect to the outlet causes a great deal of shortcircuiting in the wetland and channelization further decreases residence time. A length to widthratio based on an average of distances between inlets and the outlet is 5:8 (0.625) and residencetime for this site averages 4.4 hours. Additionally, the area of the Rio Hill wetland is less than0.95 percent of the drainage area, which is less than the minimum recommendation of 1 percent(Schueler 1992). Short circuiting is also believed to occur at the Rt. 460 site, with an inletdraining an exit ramp of I-64 within 10 meters of the outlet structure. Flow path (even for watertraveling from the farthest spaced inlet and outlet) at the Rt. 460 site is minimized as the lengthto width ratio is only 1:1. It should be noted as well that means from the Rt. 460 site include datafrom Hurricane Bertha, an extremely large event which had the effect of flushing out a lot ofdebris from the wetland.

While minimal or negative removals (export of pollutant) are indicated for the Brookewetland for TSS, OP, COD, and Zn, these figures must be viewed within the context of thesystem as a whole. A comparison of the detention basin inflow and the relatively lower wetlandinflow (detention outflow) concentrations for the Brooke wetland indicate that a significantportion of removal at the Brooke site occurs in the detention basin rather than in the wetland.While effluent from the Brooke wetland may contain higher pollutant concentrations for someparameters than the wetland inflow, the concentration is still far lower that that in the inflow tothe system, resulting in overall pollutant reductions.

SECTION II: WETLAND MODELING

BACKGROUND

Mathematical models are valuable tools for understanding the function of constructedwetlands, quantifying pollutant removal, and developing successful management strategies.Mathematical modeling of ecological systems is currently an emerging science, and little

34

attention has yet been paid to the mathematical modeling of wetlands receiving stormwaterrunoff.

Since the 1980s, a number of wetland models have been developed. These efforts rangefrom attempts to describe very specific wetland processes, to detailed models of wetlandhydrology and nutrient cycles. Many of these have been designed to simulate wetland systemsdesigned to treat wastewater; few have dealt with wetlands receiving stormwater as a primarysource. Those models that have addressed wetlands receiving stormwater runoff tend to verysite-specific and are not widely applicable.

Two primary approaches have been taken in modeling flow in wetlands: a hydrodynamicapproach (Guardo and Tomasello, 1995) and a hydrologic budget approach. While thehydrodynamic approach yields detailed flow information for a wetland system, these modelsgenerally are applied on a relatively short time scale and they neglect major components ofwetland hydrology such as evapotranspiration and infiltration. Hydrologic models, on the otherhand, are applied over longer time scales but are often formulated as black box models andyield little information concerning the internal dynamics of wetlands that affect pollutanttransport. These hydrologic models are also often applied on large spatial scales (i.e. watershedor region) and therefore provide little detailed information about specific wetland sites.

The development of water quality models for wetlands has both limitations andchallenges. A vast majority of research in wetland water quality modeling has focused on thecycling of nitrogen and phosphorus in wetlands. Detailed research has been conducted since the1970s to understand nutrient uptake by plants (Boyd, 1970, 1978; Caassen and Barber, 1976;Hearn, et al. 1991; Neilsen, 1976; Taylor, 1983). A number of conceptual models for nutrientcycling have been developed as well (Heliotis and DeWitt, 1983, J rgensen, 1986). Severalmathematical models have been applied to simulate nutrient cycles; however, they have largelybeen very site-specific. Kadlec and Hammer (1988) and J rgensen (1988) employed similarapproaches, modeling nutrients in surface water, vegetation, and three subsurface zones in greatdetail. Simpler approaches to simulating wetland mescosms include a black box phosphorusmodel (Niswander and Mitsch, 1995) and a batch reactor model (Laio, 1996).

The importance of nutrient cycles in wetlands cannot be understated, and further researchinto modeling nutrient cycles in wetlands is critical. Nitrogen and phosphorus are only two ofmany pollutants responsible for water quality impairment, however. Runoff from urban areasand highways may contain significant levels of metals and hydrocarbons that constructedwetlands may have the ability to remove or transform. A copper model by Light (1992) utilizingthe USEPA Water Quality Analysis Simulation Program (WASP) (Ambrose, 1993) model and ametal speciation model represents one of few efforts to model metal transport in wetlands asidefrom acid mine drainage sites.

35

METHODS

Modeling Criteria

The researchers evaluated three models for possible field scale application: USEPAs WASP(Ambrose, 1993), U.S. Department of Agricultures Chemical, Runoff, and Erosion fromAgricultural Management Systems model (CREAMS) (Knisel, 1980), and the VirginiaStormwater Wetland Simulation Program (VASWETS) (Liao, 1996).

The following criteria were considered in model evaluation and selection:

1. Ability to describe physical and chemical processes in a wetland system

2. Amount of data required for input, and the ability to obtain this data by measurement orestimation

3. Extent of modifications required to model a wetland system, and ability to implement thesemodifications within the framework of the existing code

The research team determined that all of the models would require extensive modifications inorder to work. For that reason, they developed a new model, the Virginia Field Scale WetlandModel (VAFSWM), part of which is based on the VASWETS conceptual model.

Modeling Approach

VASWETS models mechanisms of settling, diffusion, adsorption to plant and substrate,and vegetative uptake for a pollutant in dissolved and particulate forms in a two segment (watercolumn and substrate), two state (completely mixed and quiescent) batch reactor system.Conceptually, VAFSWM takes the kinetics portion of this batch system and applies it on a fieldscale. It adds a hydrologic subroutine to model pollutant inputs and outputs from stormwaterrunoff and to route flows through the treatment system, and a routine to model suspended solidsin the water column. Figure 24 illustrates the hydrologic balance modeled, while Figure 25illustrates the pollutant mass balance.

36

Figure 24. Wetland model - hydrologic balance

Figure 25. Pollutant mass balance approach

37

Hydrologic Balance

The movement of water in the constructed wetland is described in terms of a hydrologicbalance on the surface water segment with the volume of this segment (the water storage) as thestate variable. Subsurface inflows and outflows are assumed negligible. The governing equationfor the hydrologic balance is:

where

S = Storage [L3]Q

IN = Inflow [L3/T]

QOUT

= Outflow [L3/T] = f(S)P = Precipitation [L3/T]ET = Evapotranspiration [L3/T]EX= Exchange with subsurface Groundwater Input (EX negative) or Infiltration (EX positive)

[L3/T]

With QOUT

specified as a function of storage, the resulting equation is an ordinarydifferential equation which is then solved using a fourth order Runge Kutta integration with afifth order accuracy check to yield storage and wetland outflow.

Pollutant Mass Balance

The governing equations for pollutant mass balances for suspended solids (C1), the pollutantin the water column (C2) and the pollutant in the substrate (C3) are:

dSdt

Q Q P EXIN OUT

V dCdt

Q C Q C AC

V dCdt

Q C Q C Af C k k Vwf C k A f C f C

Vs dCdt

Af C Af C k k Vsf C k A f C f C

W IN IN OUT SW

W IN IN OUT SW PW DW DS DW

SS PS SW PW DS DS DW

1 1 1 1

2 2 2 2 2 3 2

3 3 2 3 3 2

2 3 1

2 4 1

FHG

IKJ

FHG

IKJ

b g

b g

38

where

C1 = Concentration of suspended solids [M/L3]C2 = Concentration of pollutant in water column [M/L3]C3 = Concentration of pollutant in substrate [M/L3]V

W = Volume of water column [L3]

VS = Volume of substrate [L3]

QIN

= Inflow to water column [L3/T]Q

OUT = Outflow from water column [L3/T]

vSW

= Settling velocity in water column [L/T]v

SS = Settling velocity in substrate [L/T]

k1 = First order exchange coefficient [1/T]

k2 = Adsorption coefficient to substrate [1/T]

k3 = Adsorption coefficient to plant [1/T]

k4 = Plant uptake coefficient [1/T]

fPi

= Fraction of particulate pollutant in compartment ifDi

= Fraction of dissolved pollutant in compartment i = Porosity of substrate

and

with

kp = Partition coefficient [L3/M]

The above equations assume that pollutant fluxes due to precipitation and exchange withsubsurface are minimal compared to other mechanisms. The resulting three ordinary differentialequations are solved simultaneously using a fourth order Runge-Kutta integration with a fifthorder accuracy check. Governing equations for the quiescent state are identical to the equationspresented above; however, inflows and outflows do not occur in the quiescent state and settlingvelocities are assumed to be far lower in this state. Liao (1996) utilized a two-way regression onlaboratory data to determine the duration of the completely mixed state. For the VAFSWM, thisduration must be estimated.

Model Verification

The researchers verified the VAFSWM kinetics by comparing it with analternative computational framework, VASWETS. VASWETS was chosen for comparisonbecause input data files were available and because VASWETS had been verified previously with

f k Ck C k CPi

P i

p iDi

p i

11 1

11 1

f

39

Figure 26. Model verification by comparison with VASWETS

the WASP model incorporating two-phase, two-segment kinetics (Liao, 1996). Agreementbetween VASWETS and the WASP model was excellent. Computation schemes for integrationfor the newly developed model and VASWETS are quite similar; however, the WASP modelutilizes an Eulerian step size predictor, followed by a second order Runge Kutta step sizecorrector for integration of the coupled first order ordinary differential equations. Forcomparison with VASWETS, the VAFSWMs kinetic routine was isolated by assuming thatsystem inflows and outflows were equal to zero. Precipitation, evapotranspiration, andinfiltration were also assumed to be zero. Table 12 provides a list of input parameters used in thecomparison. These input parameters were taken from VASWETS simulations for TP in a batchsystem (Liao, 1996). A comparison of model results with VASWETS results (Figure 26) showsexcellent agreement. Similar agreement was observed for a range of simulations that wereperformed by varying parameters (listed in Table 12).

40

Table 12. Model Verification Input Parameters

Parameter Description ValueC10 Initial concentration of suspended

solids25.0 mg/L

C20 Initial concentration of pollutantin water column

5.34 mg/L

C30 Initial concentration of pollutantin substrate

5.34 mg/L

k1 Transfer coefficient betweensubstrate and water column

0.0002 /day

k2 Adsorption coefficient tosubstrate

0.018 /day

k3 Adsorption coefficient tovegetation

0.058 /day

k4 vegetation uptake coefficient 0.096 /dayVW Volume of water column 7.28 10-3 m3

VS Volume of substrate 7.28 10-3 m3

Porosity 0.21vSS Setting velocity in substrate 0.44 m/dayvSW Settling velocity in water column 0.093 m/dayvSS q Settling velocity in substrate

(quiescent)0.01 m/day

vSWq Settling velocity in water column(quiescent)

0.01 m/day

kp Partition coefficient 0.021 [m3/kg]TMAX Simulation time 17 days

41

RESULTS

To illustrate application of the Virginia field scale model, a test case was devel-oped to simulate TSS and TP. This test case, which was based on a storm at Rt. 288 in July1997, was selected due to the input requirements of the model. While the amount of inputrequired for this model is far less than that of some other applicable models (Ambrose et. al.,1993; Hammer and Kadlec, 1988; and Knisel, 1980), an extremely dry summer in Virginia in1997 limited data collection for storm events. Further, the presence (until mid-summer 1997) ofthe beaver dam at Rt. 288 hindered collection of the outflow data necessary for a rigorous cali-bration. Parameters for the test case are based on data collected at Rt. 288 and on typical valuesfrom the literature. Simulation parameters are presented in Table 13. Details of parameterselection for kinetic coefficients are discussed by Liao (1996). Figure 27 shows rainfall andrunoff while Figure 28 shows TSS and TP concentrations for the inflow for the wetland simula-tion. Exchange with the subsurface (infiltration and groundwater inputs) was assumed negligiblefor this simulation. Figure 29 shows the results of this simulation. The results are illustrative ofthe capabilities of the model and were as expected.

42

Table 13. Simulation parameters for Rt. 288 test case

Parameter Description ValueC10 Initial concentration of suspended

solids5.0 mg/L

C20 Initial concentration of pollutantin water column

0.9 mg/L

C30 Initial concentration of pollutantin substrate

0.9 mg/L

k1 Transfer coefficient betweensubstrate and water column

0.0002 /day

k2 Adsorption coefficient tosubstrate

0.018/day

k3 Adsorption coefficient tovegetation

0.035/day

k4 vegetation uptake coefficient 0.055/daySW Initial volume of water column 2286 m3

SS Volume of substrate 4572 m3

SDEAD Volume of dead storage 2286 m3

ET Evapotranspiration rate 2.20 mm/dayA Area of wetland (plan view) 15000 m2

Porosity 0.30vSS Setting velocity in substrate 0.45 m/dayvSW Settling velocity in water column 0.15 m/dayvSS q Settling velocity in substrate

(quiescent)0.01 m/day

vSWq Settling velocity in water column(quiescent)

0.01 m/day

kp Partition coefficient 0.00202 m3/kgCD Discharge coefficient 0.9L Weir length 0.9144 mTMAX Simulation time 2.5 days

Figure 27. Rainfall and runoff for wetland simulation

Figure 28. Influent pollutant concentrations for wetland simulation

43

To illustrate the ability of the model to simulate wetland water quality over anextended period of time, a simulation was also conducted for the Rt. 288 site for 1996. Kineticparameters were the same as those used for the event simulation. A continuous precipitationrecord from the Richmond International Airport was used for input, and inflows that were notmonitored at the site were determined based on a regression of rainfall and average daily flowfrom monitored events. Evapotranspiration rates used for the simulation were taken from thirty-year average monthly evapotranspiration rates as calculated by the Thornthwaite method. EMCsfrom monitoring data from 1996 were utilized for input for monitored storms, while a mean ofEMCs from all events monitored at the site were used as inflow concentrations for storms thatwere not monitored. Again, the necessity of estimating much of the input data and the lack ofsufficient site data prevent formal calibration. Results of the simulation are shown in Figure 30.

44

Figure 29. Event simulation results

Figure 30. 1996 Simulation results

DISCUSSION

While the inability to formally calibrate the model at this time prohibits inference ofactual wetland performance from modeling results, the mass removal efficiencies calculated bythe model for this simulation of 48 percent for TSS and 24.5 percent for TP are in the range ofremoval efficiencies reported in the literature and observed in the field at the Rt. 288 site.Removal efficiencies of 79 percent for TSS and 47 percent for TP for the extended simulationalso agree well with values from the literature.

Though a design storm approach is commonly used for the design of stormwaterfacilities for the control of runoff quantity, such an approach is not as applicable for the design ofa wetland to improve water quality. Long-term wetland performance is affected by storage andtransformation of pollutants within the wetland and by seasonal changes in nutrient uptake,vegetation growth, senescence, and nutrient release. Vegetative uptake (and release) of pollutantsand other significant processes such as diffusion occur at rates that are orders of magnitudeslower than more rapid processes such as sedimentation. A design approach based on a singleevent fails to account for the potentially significant impact of these less rapid processes on longterm removal efficiency. Continuing modifications to the model include the incorporation ofmore sophisticated uptake and release kinetics to more accurately model the seasonal variationsof these processes for simulations of long term performance.

45

SECTION III: GEOGRAPHIC INFORMATION SYSTEM DEVELOPMENT

BACKGROUND

VDOT and other state departments of transportation face a significant challengein deciding how best to maintain their state inventories of mitigated wetlands. Specific informa-tion pertaining to many of Virginias mitigated wetlands has become increasingly difficult tolocate and maintain due to employee turnover, the disperse locations of the sites, and the elapsedtime since initial site creation. Monitoring requirements placed on VDOT by the US ArmyCorps of Engineers continue to increase as well.

A geographic information system (GIS)-based inventory system is a usefulmethod of storing information on existing sites, as well as providing needed data for futurewetland planning. Proper site selection is one of the most important factors determining thesuccess of constructed wetlands: A well-designed GIS allows planners to factor in numerousspatial data sets prior to committing any resources to site construction.

Since modeling tools typically lack sufficiently flexible spatial analytic components, GIScan also be used to provide spatially varied data for selected stormwater management models. Alink between the data residing in a wetland GIS and the selected stormwater management modelenables users to input spatially varied data, enhancing the accuracy of model simulations.

The GIS developed as a part of this study has three applications or modes:

1. Constructed wetland inventory. The inventory stores information on existing VDOT-created mitigation sites the agency monitors.

2. Wetland site selection guide. The guide serves as a tool to help locate new mitigationsites.

3. Arc/INFO GIS-stormwater management model interface. This interface provides alink between data stored in the GIS and a stormwater management model.

Each of these modes could be useful to personnel responsible for wetland mitiga-tion in and of themselves; however, VDOTs Environmental Division is in the early stages of thecreation of a division-wide GIS. The above three modes may or may not become part of thedivisions GIS, depending on the type of system that is ultimately developed. At the very least,the GIS created as a part of this study should serve as a learning tool for further GIS developmentin the Environmental Division.

METHODS

Constructed Wetland Inventory

A mitigation site database was developed using PC Arc/INFO and ArcView 3.0.To create a point coverage representing VDOT created wetland sites, latitude/longitude coordi-

46

nate pairs for all existing mitigation sites were taken from project files, USGS quad sheets, and/or global positioning system (GPS) reading using a Trimble GeoExplorer receiver. All GPSreadings were differentially corrected using a public base station located in Charlottesville. Forsome of the larger sites, GPS was used to delineate the entire wetland boundary. GPS files wereexported as Arc/INFO coverages using Pathfinder Office GPS software.

In addition to the point coverage created for the sites, all existing attributes for thewetland sites, previously stored in a rudimentary database by VDOTs Environmental Division,were transferred to the attribute table of the coverage. Attributes added include: project name,project number, permit number, watershed, impact type, impact function, impact size, mitigationtype, mitigation function, mitigation size, monitoring due dates, and monitoring comments.Once coverage development was complete in Arc/INFO, a shape file was created in ArcView,thereby allowing additional edits to be made within the ArcView environment. County and stateboundary themes were added to give the user a spatial frame of reference when using the wetlandsite point theme. Additional information, such as digital photographs, aerial infrared photo-graphs, GPS derived delineations, and detailed maps indicating major areas of vegetation, inlets,outlets, etc., were also hot-linked to specific sites when available.

The research team used Avenue programming language to slightly amend thestandard ArcView graphical user interface (GUI) for the constructed wetland inventory. In orderto simplify the standard ArcView interface, some of the default options and tools normallyavailable to the user were eliminated. The simplified interface reduces the chances that aninexperienced user will inadvertently alter the database, while allowing more experienced per-sonnel sufficient access to complete more complex queries and analyses, and to update thedatabase as changes to the wetlands occur.

Site Selection

To aid VDOT personnel in the selection of potential sites for newly createdwetlands, the second phase of the GIS construction included the development of a site selectionmode. Numerous factors are considered in a siting selection; the research team conducted aliterature search and conducted interviews with VDOT wetland personnel in order to identify thespecific data sets necessary for this process.

A total of 13 data sets or layers of information were identified as being useful. Fromthese, a weighting system was used to select out the most important data sets. Following theidentification process, data were acquired from a variety of sources at the largest scale (highestresolution) possible.

The Staunton District was chosen as the study area for this portion of the GIS due prima-rily to the availability of data for this area of the state. The coverages were created within Arc/INFO and transferred to ArcView for viewing, query, and analysis. A specific sequence ofqueries was developed to aid the user in the selection process of potential sites. Additionalfunctions were added to the default ArcView GUI to assist the user in carrying out necessarysteps for site selection.

47

48

Modeling Linkage

Modeling tools typically lack sufficiently flexible spatial analytic components. To ad-dress this deficit, the researchers searched for an available link between existing stormwatermodels and GIS, thereby allowing the user to take advantage of the spatial analysis capabilities ofthe GIS. They conducted a literature search to determine the availability of existing stormwatermodel-GIS interfaces. Those interfaces deemed potentially suitable were acquired and tested forapplicability for VDOT using sample data sets. Factors that were considered when characteriz-ing the interfaces included: model usage, cost, platform requirements, operational difficulty, anddata requirements. A single interface was selected and further tested using data for a smallwatershed in Rappahannock County. Finally, the researchers identified modifications that wouldimprove the interface.

RESULTS


Figures 31 and 32 are screen captures of the GIS-constructed wetland inventory mode.All available information for the over 220 created wetland sites is available to the user by click-ing on the point representing the site, by viewing the entire attribute table for the theme, or bysubmitting a user-defined query.

Figure 31. Constructed wetland mode showing attribute table and infrared aerial photograph

Figure 32. Constructed wetland mode showing site attributes. GPS delineation and ground photographs

49

The researchers identified 13 data sets that are useful for site selection: hydrologic unitcodes, streams and rivers, groundwater hydrology, flooding potential, land use, soil type, vegeta-tion cover, elevation (topology), transportation, land cost, National Wetlands Inventory, waterintake/discharge permit points, and endangered species locations. The most critical of these 13data sets were prioritized using a weighting scheme devised by Daigle (1996). After applyingthis weighting scheme, the data sets determined to be necessary for future mitigation site selec-tion included: soils, water bodies, streams, roads, land use, hydrologic unit codes, and elevationdata. All of these coverages were obtained for the Staunton District, most at a scale of1:100,000. Most were taken from the Southern Appalachian Assessment GIS Data Base (1996).Higher resolution data were requested from various sources, but could not be obtained for all thecounties making up the district. Additional information (such as vegetation cover, endangeredspecies habitat, land cost, and groundwater hydrology) would be extremely useful for this modeof the GIS, but were only available for relatively small, discontinuous areas throughout the state.Figure 33 depicts the site selection overlay procedure.

From the Site Selection mode GUI, the user can issue the commands necessary to identifysites meeting predetermined criteria. The recommended order of data display and query is asfollows:

1. Determine hydrologic unit code (HUC) for which wetlands will be mitigated

2. Select this HUC with the selection tool and zoom in to the extent of this using zoom to selec-tion tool (Figure 34)

3. Select areas containing the desired land use(s) using the query builder (Figure 35)

4. Select areas meeting soil type requirements from previously selected set

5. Draw streams and roads coverages (Figure 36)

6. Create buffers of a specified distance for both streams and roads using buffer tool

7. Visually determine intersections between buffered sections and previously selectedpolygons

8. Determine sites that meet topography requirements from this set by intersecting withelevation coverage or by making elevation theme active and requesting elevationinformation for each of the potential polygons individually.

50

Figure 33. Site selection coverage overlay procedure

Figure 34. Site selection mode with hydrologic unit codes highlighted

Figure 35. Site selection mode with land use and query builder displayed

51

Figure 36. Site selection mode showing roads, streams, and land use intersection

Maximum slope characteristics can be derived from the elevation coverage by way ofanalysis tools provided in the Spatial Analyst extension in ArcView. This extension, whileproviding various other hydrologic functions, allows for the derivation of a watershed, againfrom the elevation coverage. If the user would prefer to have the replacement site located withinthe same local watershed (as opposed to the same hydrologic unit code), the watershed can bedelineated, the area calculated, and the polygon selected as the initial steps in place of the previ-ously described step 1.

Modeling Linkage

Numerous hydrologic model GIS interfaces have been developed over the past five toseven years. The majority of the earlier interfaces were what is referred to as loosely coupledinterfaces, meaning that the GIS simply output data in a format that was readable by the model.Closely coupled models, those that pass data between the model and the GIS via memory-resi-dent data models as opposed to external files, are becoming more common (Haddock andJankowski, 1993; Liao and Tim, 1997). While significantly more difficult to develop, closelycoupled interfaces are much faster and more efficient for the user.

52

Four stormwater model-GIS interfaces were identified in the literature search. Theinterfaces include Watershed Modeling System (WMS), Agricultural Nonpoint Source Model(AGNPS), Better Assessment Science Integrating Point and Nonpoint Sources (BASINS), andArcView Sand (AVsand). The results of the model characterization can be found in Table 14.Interfaces that connect other GIS packages such as Geographical Resource Analysis SupportSystem (GRASS) and AGNPS do exist, but were not investigated in this study (Osmond et al.,1997).

The AGNPS interface, a closely-coupled interface developed by Hsiu-Hua Liao and U.S.Tim at Iowa State University, was tested for potential VDOT application. AGNPS had severaladvantages over the other interfaces, including cost and the fact that it linked Arc/INFO toAGNPS 5.0. Arc/INFO is a fully functional GIS, providing the user with a much greater set ofspatial analysis functions than most desktop GIS packages, such as ArcView, to which theother interfaces are linked. AGNPS is a powerful, single event, distributed parameter model thatsimulates nonpoint pollution from watersheds. The model can be used to predict runoff volume,eroded and delivered sediment, chemical oxygen demand, nitrogen concentrations, and phospho-rus concentrations in the runoff. The AGNPS model was developed by the Agricultural ResearchService in 1987 (Young et al., 1987). The AGNPS interface will compute nonpoint source loads(and some point source loads) for a number of pollutants and a wide range of watershed sizes.The interface therefore may be used for a general or detailed water quality analysis. Figures 37-40 show the AGNPS interface for each of the programs four modules.

Table 14. GIS-Stormwater Model Interface Characterization

Name AnalysisEmphasis

ApproximateCost

OperatingSystem

GIS - ModelLinkage

Difficulty Notes

WMS waterquantity

$750Windows

95Windows

NTUNIX

ArcView w/HEC-1

NFFTR-20

RationalMethod

lowExcellent fordelineation andextraction ofwatershedcharacteristics

BASINS waterquality

$0 Windows95

ArcView w/HSPF

QUAL2ETOXIROUTE

moderate Analysis done atcounty level orsmaller scale

AGNPS waterquality

$0 UNIX Arc/INFO w/AGNPS

high UNIX skillsnecessary

AVsand waterquality $4,400

Windows95

WindowsNT

UNIX

ArcView w/SWMMSAND

moderateMore userfriendly

53

Figure 37. AGNPS-Arc/INFO data generation module

54

Figure 38. Input file creation module

55

Figure 39. AGNPS Execution module

56

Figure 40. AGNPS-Arc/INFO ouput file extraction module

DISCUSSION


The research team selected ArcView, Environmental Systems Research Institutes desktopGIS package, as the software for this application primarily because of its compatibility with Arc/INFO and because it is one of the more user friendly GIS packages available. It was assumedthat the end user of the Constructed Wetland Inventory would not necessarily have extensiveformal training in GIS, and therefore would not need a more sophisticated/complicated package.Additionally, since it was assumed that any changes or updates to the system would be done inArcView, all Arc/INFO coverages used were converted to ArcView shape files (shp). Thiseliminates any need for users to make changes in Arc/INFO. All available information for thesites is included in the GIS. Additional information on new or existing sites will be added tothe system as it becomes available. If not updated consistently, the system will not serve itsintended purpose and will become increasingly difficult to manage.

Potentially, VDOT could utilize the GIS developed for the mitigation for three types oftasks: monitoring, maintenance/remedial action, and long term management. Monitoring of

57

mitigation sites ensures that the hydrology and planting criteria established in the permit condi-tions are being met. The US Army Corps of Engineers requires that most emergent sites bemonitored for five years; forested sites may require monitoring for 10 or even 20 years. Storingall data digitally will lessen the burden of data management. The photographs stored in thesystem will allow for direct comparison of vegetative success at specific locations from year toyear. Digitally mapping the wetland delineations, specific habitats, and areas of major vegetationdominance will allow for quantitative spatial analysis between sites and within the same site overtime.

The GIS database will enhance maintenance and remedial actions required by providing atracking mechanism for work conducted at the sites (e.g., a specific area may require replantingseveral times). As data are collected for the sites and added to the system over the years, it willbe possible to carry out trend analyses. The GIS will aid long term management of the sites too.Ideally, this system will be linked with the Environmental Divisions GIS and ultimately toVDOTs department-wide GIS, making the wetland information available to environmentalpersonnel outside of Aquatic Ecology as well as to other VDOT divisions. By having the spatialinformation available, divisions which would not normally know the location of the wetlandsites, will be able to avoid them when conducting unrelated work tasks (e.g., maintenance wouldknow where to avoid spraying herbicides).

Site Selection

Proper placement of newly constructed wetlands is a critical component of their ultimatesuccess. Placement of the mitigation site in the same hydrologic unit code (HUC) is, in mostcases, required (Daigle, 1996). Other hydrologic criteria can be derived from the streams andwater body coverage as well as the elevation or topography coverages. The GIS land use cover-age provides land use and land cover information based on the Anderson classification systemtaken to the level 2 characterization (Hermann, 1996). Desired proximity to roads and streamscan be specified using these coverages along with the buffer tool.

The site selection mode of the GIS, like most any information system, is only as good asthe data that comprises it. A great deal of time was spent trying to gather the data required forthis mode. The data sets that were obtained are sufficient to allow the user to identify potentialareas to investigate further; possibly more importantly, the user can quickly eliminate areas forconsideration. Additional information (such as land cost and groundwater hydrology), if it werereadily available, would greatly benefit this mode of the GIS. Also, larger scale (higher resolu-tion) data (such as soils) would also be very helpful. There is a tradeoff when gathering this kindof data, however: Some forms of data acquisition (i.e., digitizing higher resolution soils data)could easily offset any time saving benefits of the system. As the Environmental Divisions GISbegins to develop, these additional, high-resolution data sets should become more readily avail-able.

Modeling Linkage

It was originally envisioned that a GIS-stormwater model interface would be developedas a part of this project. During the literature search, however, it became evident that severalinterfaces already existed for a variety of GIS packages and stormwater models. It also became

58

clear that developing a closely-coupled interface would be well beyond the scope of the originalproject. The AGNPS interface provides all the functionality needed and is public domain. FromVDOTs perspective, the most significant drawback to this interface is the fact that it currentlyonly runs on a UNIX platform, which requires the user to run UNIX Arc/INFO and AGNPS forUNIX. The majority of VDOT environmental personnel do not have easy access to these pro-grams and are not proficient with the UNIX operating system. Some of the other PC-basedinterfaces that have been developed link PC Arc/INFO to stormwater models, but these arelimited by the reduced capabilities of a PC-based GIS. This constraint alone can severely limitthe size of the watershed being modeled and/or the resolution of the data sets used for modelinput. The developer of the AGNPS interface indicated that both the interface and AGNPS couldbe compiled to run on Windows NT (Liao, 1997). The interface was written in C code, while theactual AGNPS program was written in FORTRAN code. Therefore, the AGNPS interface andprogram should be compilable within Windows NT. The Windows NT version of Arc/INFOclosely resembles that of UNIX Arc/INFO in overall functionality.

The AGNPS interface will significantly reduce the time requirement for constructingAGNPS input and output files. Before construction of the interface, users had to manually entermore than 20 inputs for each cell of grids covering each study area. Hence, data entry wasextremely time consuming when modeling watersheds with many cells. With the interface,inputs for each cell are automatically derived for grids containing up to 32,000 cells (Liao, 1997).

Even with this reduced data entry time, however, considerable time should be allocatedfor preparing the GIS coverages. AGNPS is very data-intensive when a high level of accuracy isdesired. For example, when modeling relatively small watersheds (1.5 15 km), the use of high-resolution data is imperative for accurate results. Consequently, data at or greater than 1:24,000scale is desirable for these watersheds. Another time- consuming aspect of the AGNPS interfaceis the checking of overland flow paths derived from a triangular irregular network (TIN), one ofthe GIS coverages required by the interface. Determining the overland flow path for every cell ina grid is cumbersome when working with coverages composed of a relatively large number ofcells.

CONCLUSIONS

1) Mitigated wetlands receiving highway runoff may be as effective as conventional BMPs atimproving the quality and controlling the quantity of highway runoff. Peak reductions inexcess of 40 percent were observed, with attenuation of greater than 90 percent for a systemcombining a detention basin and a mitigated wetland in series. Aver

FINAL REPORT CONSTRUCTED WETLANDS FOR STORMWATER ... Wetland… · FINAL REPORT CONSTRUCTED...

Documents

Transcript of FINAL REPORT CONSTRUCTED WETLANDS FOR STORMWATER ... Wetland… · FINAL REPORT CONSTRUCTED...